Skip to content
llm-speed

Measurement wanted · request DR-028 · opened 2026-11-19

Hy4 Flash on 128GB Spark: not measured yet

We don't have a Hy4 Flash speed run on 128GB Spark yet, so there is no tokens-per-second number here. People asked for it, and this page is where the result will appear.

We haven't measured its speed yet. Not with headroom, if you load the full 125.73 GiB file. That leaves about 2.3 GiB on a 128 GiB machine, which is not enough for the operating system and a KV cache unless you offload. We have not measured offload. The 156.79 GiB Q4_K_M default does not fit in 128 GiB at all. Not in stock llama.cpp, Ollama or LM Studio yet. llama.cpp#32517 (https://github.com/ggml-org/llama.cpp/pull/32517) is open and still a draft (checked 19 November 2026). Check back. File-size fit only: https://llm-speed.com/m/hy4-flash

What's being requested

Model
Hy4 Flash · AngelSlim/Hy4-flash-GGUF
Hardware
128GB Spark (the CLI records the machine's total RAM, rounded to 8 GB, not a container limit, so only runs checked for this RAM case count)
Also requested
Apple M5 Max 128GB, 4× RTX 3090

Runs we already have

No Hy4 Flash runs on any hardware yet.

Get an alert

Email alerts for this page are switching on soon. Until then, follow @llmspeed on X, where new runs are posted.

All requested runs